مدلسازی تاثیر افزایش سن بر هزینه های تامین اجتماعی

In this paper, we survey the features of different approaches available in the literature used to study the effects of population aging on Social Security expenditures. We comment on the weaknesses and strengths of each of them, and perform a quantitative analysis by comparing the results they imply in the particular case of the Spanish economy. Finally, we highlight some elements of the modelling strategies on which more evidence is needed for a correct evaluation of the problem at hand.

During the forthcoming decades in most European countries the demographic scenario will dramatically change. The fall in fertility experienced during the last quarter of the XXth Century and the continuation of the rise in longevity will lead to a significant increase in the proportion of the older population. Thus, Social Security programmes, whose expenditures are very much determined by the size of the older population, will come increasingly under financial stress. At least since the early 1980s there have been many studies trying to quantify the rise in Social Security expenditures as a result of population aging (see, for instance, World Bank, 1994 and Roseveare et al., 1996). Nowadays, the task continues, as many political institutions are concerned by the budgetary implications of demographic changes. Over the years, the methodologies used to yield some quantitative forecasts of the likely evolution of Social Security expenditures have been improved and, nowadays there is a menu of alternative approaches to perform this task. In this paper, we survey the features of different approaches available in the literature used to study the effects of the aging of the population on the sustainability of the social security system. We group them into three categories that we label as: i) aggregate accounting, ii) individual life-cycle profiles, and iii) general equilibrium models. We highlight the weakness and strengths of each of them, and compare their predictions about the evolution of Social Security expenditures for the Spanish case. As it will be made clear, there are some crucial elements of these modelling strategies on which more evidence is needed for a correct evaluation of the problem at hand. Notwithstanding, all of the quantitative exercises we perform conclude that the points of GDP which, under the current pension schemes, would have to be devoted to expenditures in Social Security are significantly higher than current expenditures. The survey is structured in four more sections. 2, 3 and 4 present the main features of the three approaches we analyze (aggregate accounting, individual life-cycle profiles, and general equilibrium models, considering both a closed economy and a small open economy). Section 5 comments on the results from each approach when they are applied to study the Spanish situation. Finally, Section 6 contains some concluding remarks, mainly addressed to highlight some elements on which more microeconomic evidence is needed to improve the modelling of the impact of population aging on Social Security expenditures.

In this final section we highlight some the features of the alternative approaches for modelling pension expenditures that we think are important in order to properly assess the problem at hand, and the need of more empirical evidence concerning theses issues. A first important issue mentioned in the former sections is the difference between the general equilibrium approach and the others in accounting for possibly important feedback effects arising, for instance, from changes in taxes induced by demographic change. We think that a properly specified behavioral model of the joint determination of savings and labor supply in a life-cycle context is needed so as to properly address the way in which the rules used to compute pension benefits, the demographic change and, in general, policy reforms may or may not affect the finances of the social security system. However, as utility-based general equilibrium models tend to enter very rapidly into the “curse of dimensionality” it may be reasonable to abstract from changes in prices, when assessing the effects of demographic change on pension expenditures (in particular, if considering a small open economy). In any case, although in general the direction of the effect is not controversial and in fact it is a general modelling strategy, the importance of demographic changes for asset returns and financial markets is still a disputable issue, on which more empirical evidence is needed. In this sense, Yoo (1994) presents time series estimates of the relationship between asset returns and the age distribution and finds a statistically significant negative correlation between the fraction of the U.S. population aged 45 to 54 and the returns of several types of assets. In contrast, more recently Poterba (2004) suggests that the results are in general sensitive to choices about the econometric specification. A second relevant issue is the consideration of enough heterogeneity among individuals of the same generation. Since the pension formula depends on the labor market history of individuals, a suitable model should in principle account for some heterogeneity along several dimensions, that could be modelled either deterministically or stochastically. For instance, heterogeneity in terms of education could in principle be modelled in a deterministic way, upon which there is no uncertainty. However, individuals should also face enough labor market risk and choose optimally labor supply along the life-cycle so as to properly study the way in which the features of the social security system affect that decision. For instance Lazear (1985) suggested that defined-contribution pension systems are more efficient in the sense that there is a close link between contributions and benefits that is taken fully into account by individuals when deciding how much work. In addition it is likely that the incentives created by the penalties for early retirement will be a function of accumulated financial wealth and social security entitlements. Jimenez (2005) has carefully study the monetary incentives for early retirement in Spain, although the estimation performed abstracts from financial wealth accumulation decisions. In addition, although his approach can incorporate very detailed sources of heterogeneity and consequently of explaining factors, they may not be very useful to address the welfare consequences of policy changes and the estimated relationships may not be necessarily stable. In addition, general equilibrium models (Sanchez-Martin, 2001) do not find a significant effect of changes in the number of years used to compute pension benefits on the retirement pattern of older workers. Consequently, more microeconomic empirical evidence concerning these issues is clearly needed in order to shed some light on the topic.